OpenAI, Anthropic, and Google DeepMind have been collaborating for several weeks to develop a framework for artificial intelligence safety. Chris Lehane, global head of policy at OpenAI, confirmed the joint efforts during a press conference in Washington.
The initiative follows a proposal by Anthropic CEO Dario Amodei to slow the development of frontier models to mitigate potential risks. OpenAI CEO Sam Altman, Elon Musk, and Google’s Demis Hassabis have expressed support for this cooperation.
Legal experts and industry figures, including Cohere CEO Aidan Gomez, have raised concerns that such coordination could violate U.S. antitrust laws by restricting output or creating a cartel. Gomez argued that these safety measures might be used to preserve the market dominance of top labs.
OpenAI is currently supporting two legislative proposals, including the FRONTIER Act, which mandates independent audits for new AI models. Additionally, the company backs a bill being developed by senators John Thune, Ted Cruz, and Amy Klobuchar to grant the Commerce Secretary authority over model safety audits.
U.S. President Donald Trump has publicly rejected calls for slowing AI development, citing concerns that it could allow China to gain an advantage. His advisor David Sacks has also characterized fears regarding existential AI threats as exaggerated.
